Abstract

This paper presents a real-world reinforcement learning agent that uses non-verbal robot behaviors to facilitate synchronized movement between strangers during ice-breaking scenarios, exploring how embodied timing cues can ease initial social interaction.